R/get_synonym.R

In AlexisMayer/toolbox:

#' This function allows you to get synonym of a word thanks to Thesaurus.com.
#'
#' @param word Vecteur contenant le(s) mots recherché(s)
#' @param langue Choix de la langue ("en" ou "fr")
#' @param most_recommended Doit-on seulement chercher les synonymes les plus recommandés ou une liste plus exostives ? (TRUE ou FALSE)
#' @export
#' @examples
#' get_synonym("cat")
#'
#' Return larger list of synonym
#' get_synonym("cat", most_recommended = FALSE)
#' # For multiple research
#' lapply(c("cat","dog"), get_synonym)

get_synonym = function(word, langue, most_recommended = TRUE) {
   # Depedencies
   tidysession(lst = c("dplyr", "rvest", "stringr"))
   
   if (langue == "en") {
      # Get http path
      page = paste0("https://www.thesaurus.com/browse/", word)
      # Message
      cat("Recherche: ", paste(word), "\n")
      
      # Load web page
      get = try(read_html(page) %>%
                   html_nodes(".css-0") %>%
                   xml_contents())
      
      # Grep
      syn = get[4] %>%
         as.character() %>%
         strsplit("data-emotion-css") %>%
         unlist() %>%
         .[3] %>%
         strsplit("browse") %>%
         unlist() %>%
         .[-1] %>%
         str_replace_all("[[:punct:]]", " ") %>%
         trimws() %>%
         word(1) %>%
         paste(collapse = ", ")
      
      if (most_recommended == FALSE) {
         syn2 = get[4] %>%
            as.character() %>%
            strsplit("data-emotion-css") %>%
            unlist() %>%
            .[4] %>%
            strsplit("browse") %>%
            unlist() %>%
            .[-1] %>%
            str_replace_all("[[:punct:]]", " ") %>%
            trimws() %>%
            word(1) %>%
            paste(collapse = ", ")
         
         syn = paste(syn, syn2, sep = ", ")
      }
      
      # Print results
      cat("Résultats: ", paste(syn), "\n")
      
      # Return final list
      return(syn)
   }
   else if (langue == "fr") {
      # Get http path
      page = paste0("http://www.synonymes.com/synonyme.php?mot=", word)
      
      # Load web page
      get = try(read_html(page) %>%
                   html_nodes(".defbox") %>%
                   xml_contents(),
                silent = TRUE)
      
      # Grep
      syn = get[3] %>%
         as.character() %>%
         str_replace_all("[[:punct:]]+", " ") %>%
         str_extract_all(">(.*?)<") %>%
         unlist() %>%
         str_remove_all("<|>") %>%
         str_replace_all("\\s+", "") %>%
         enframe() %>%
         filter(!value == "") %>%
         select(-name)
      
      # Return final list
      return(syn)
   }
}